D E V E L O P M E N T OF A L E A F SPRING U - B O L T L O A D T R A N S D U C E R : P A R T OF A N

ONBOARD WEIGHING S Y S T E M FOR OFF-HIGHWAY L O G T R U C K S

by

MITHUN K A R U N A K A R SHETTY

B.E., The University of Mumbai, 2002

A THESIS S U M I T T E D IN P A R T I A L F U L F I L M E N T OF
T H E R E Q U I R E M E N T S FOR T H E D E G R E E OF

M A S T E R OF A P P L I E D S C I E N C E

in

T H E F A C U L T Y OF G R A D U A T E S T U D I E S

(Forestry)

T H E U N I V E R S I T Y OF B R I T I S H C O L U M B I A

M a y 2006

© Mithun Karunakar Shetty, 2006

 ABSTRACT

This thesis was motivated by the current concern of brake failure in off-highway log

trucks descending steep grades. In order to utilise a guideline being developed for the

prediction of safe maximum grades for descent under a range of truck payloads, it is

necessary to measure axle weights during loading. A background review found that there

are no commercially available on-board weighing systems that can be retrofitted to the

drive axles of an off-highway tractor. Therefore, an investigation into the development of

an on-board weighing system for the off-highway log trucks was initiated.

This research was divided into two stages: preliminary strain measurement with a

loaded off-highway tractor, and finite element modelling of a U-bolt from the tractor's leaf

spring suspension. A preliminary measurement test was carried out to identify potential

suspension components that could act as load transducers for measuring axle weight. The

preliminary results showed that incremental strain at two locations on the U-bolt varied

linearly with payload, for an incremental load of 22.5 k N . Finite element modelling of the

U-bolt was carried out to predict the maximum incremental strain occurring on the U-bolt

surface. The model was calibrated with the measured data and a sensitivity analysis was

done on key modelling parameters to determine the most suitable level of leaf spring block

length, preload and U-bolt-to-leaf spring friction coefficient. Incremental strain on the top

of the curved portion of the U-bolt was found to be relatively consistent and close to the

maximum level of incremental strain and is recommended as a preferred position for strain

gauging.

ii
 T A B L E OF CONTENTS

ABSTRACT : ii
TABLE OF CONTENTS iii
LIST OF TABLES '. iv
LIST OF FIGURES v
ACKNOWLEDGEMENTS vii
CHAPTER 1 INTRODUCTION 1
1.1 Need for investigation 1
1.2 Background review 4
1.3 Objectives 10
1.4 Organisation of the thesis 11
CHAPTER 2 PRELIMINARY STRAIN MEASUREMENTS 12
2.1 Sensor location on the HDX suspension 12
2.2 Instruments used 13
2.2 Methodology 14
2.3 Results and Discussion • 15
CHAPTER 3 FINITE ELEMENT MODEL CONSTRUCTION 21
3.1 Overview 21
3.2 Background of U-bolt modelling 22
3.3 FEM description 25
3.4 Meshing of the FEM 29
3.5 Boundary conditions 30
3.6 Preload modelling 31
3.7 External load 32
3.8 Contact modelling 32
3.9 Postprocessing 34
CHAPTER 4 FINITE ELEMENT MODELLING RESULTS 35
4.1 Overview 35
4.2 Sensitivity Analysis 35
4.2.1 Leaf spring block length 36
4.2.2 Preload 38
4.2.3 Coefficient of friction 41
4.3 Analysis of incremental strain .42
4.4 Possible causes of bending in the U-bolt shank 47
4.5 Limitations of the empirical measurements and finite element modelling 52
CHAPTER 5 SUMMARY, CONCLUSION AND RECOMMENDATION 54
REFERENCES 56
APPENDIX I .' — 58
APPENDIX II 60

iii
 LIST OF T A B L E S

Table 2-1: Estimated weight of water added and resulting drive axle group weights 16

Table 3-1: FEM component physical properties 29

Table 4-1: Run sequence inputs to evaluate leaf spring block length.... 37

Table 4-2: Run scenario for the different levels of preload and coefficient offrictionbetween the curved
portion of U-bolt and leaf spring 39

iv
 LIST OF FIGURES

Figure 1.1: Typical off-highway log truck 1

Figure 1.2: Off-highway log truck axle weights used in forest road bridge design 2

Figure 1.3: Walking beam load transducer 4

Figure 1.4: Log bunk assembly on a highway log truck 5

Figure 1.5: Beam load cell mounted on a highway log truck 5

Figure 1.6: Trailer suspension assembly with SI beam load cell 6

Figure 1.7: Bunk assembly on an off-highway log truck 6

Figure 1.8: Leaf spring suspension of an off-highway tractor 7

Figure 1.9: Load cells mounted between a drive axle and its leaf spring suspension 9

Figure 2.1: Strain gauge locations 13

Figure 2.2: Measuring drive axle weights with portable pad scales 15

Figure 2.3: Incremental leaf spring strain response to unloading the tractor with strain gauges applied when
tractor was loaded 17

Figure 2.4: Incremental axial strain measured in two locations on a leaf spring U-bolt during unloading
with strain gauges applied when tractor was loaded 18

Figure 2.5: Incremental axial strain at location 1 on the suspension U-bolt in response to an averaged drive
axle group weight 20

Figure 3.1: FE modelling procedure for the U-bolt analysis 22

Figure 3.2: Finite element modelling of U-bolted assembly 23

Figure 3.3: FEM of a U-bolted leaf spring assembly... 24

Figure 3.4: Schematic of the U-bolt assembly 26

Figure 3.5: 3-D 10-node tetrahedral structural solid... 27

Figure 3.6: Location of contact elements 28

Figure 3.7: Surface-to-surface contact elements 28

V
Figure 3.8: FEM of the U-bolt assembly 30

Figure 3.9: Boundary condition 31

Figure 4.1: Incremental strain at gauge location 1 on the U-bolt for various leaf spring block lengths 37

Figure 4.2: Incremental strain at gauge location 2 on the U-bolt for various leaf spring block lengths 38

Figure 4.3: Incremental strain at gauge location 1 on the U-bolt for various preloads and coefficients of
friction (u) between the U-bolt and the leaf spring 40

Figure 4.4: Incremental strain at gauge location 2 on the U-bolt for various preloads and coefficients of
friction (u) between the U-bolt and the leaf spring 40

Figure 4.5: Simulated incremental strains at locations 1 and 2 on the U-bolt for different U-bolt-to-leaf

spring coefficients of friction 42

Figure 4.6: Distribution of incremental strain along the outer edge of U-bolt surface 43

Figure 4.7: Incremental strain distribution along the inner and outer surfaces of the U-bolt shank 44

Figure 4.8: Bending strain distribution along the outer surface of the U-bolt shank 45

Figure 4.9: Recommended gauge location on the U-bolt 47

Figure 4.10: Bending in the U-bolt 48

Figure 4.11: Geometry of two-dimensional model 49

Figure 4.12: Contact elements in two cases 50

Figure 4.13: Deformed shape for case 1 50

Figure 4.14: Deformed shape for case 2 51

Figure 4.15: Defining U-bolt Volume in FEM 52

Figure A.I: Strain transformation along the curved portion of the U-bolt 58

Figure A.II: Slip-on water tank dimensions 60

vi
 ACKNOWLEDGEMENTS

I would like to express my sincere gratitude and thanks to my supervisor Dr. K e v i n C.

Lyons for financial support and his invaluable help over the whole course of my studies.

M y special thanks to M r . A l l a n Bradley of Forest Research Institute of Canada ( F E R I C ) for

his invaluable professional advice, time and effort spend on proofreading this report. I

would also like to thanks his family for their support during my stay in Canada.

My sincere thanks to M r . Marv Clark of F E R I C for providing constructive

feedback, financial assistance and instruments used in the preliminary testing. I would also

like to thank Dr. John Nelson for participating on the committee for this thesis. I would

like to acknowledge M r . A l l a n Waugh of Hayes Forest Services (Hayes) for providing a

truck and shop facilities for the preliminary testing, and M r . Jeff Layfield of Hayes for

providing useful information required in this work. Thanks to the Instrumentation Lab,

Dept. of Mechanical Engineering, U B C for providing instruments used in the preliminary

testing. I would like to thank M r . Rob Jokai and M r . Seamus Parker of F E R I C for their

help with using FERIC-supplied instruments.

I owe my parents a special gratitude for their love, understanding and support

during the course of the study.

M.K.S.

vii
 CHAPTER 1

INTRODUCTION

1.1 Need for investigation

Heavy-duty off-highway log trucks, as shown i n Figure 1.1, have a very limited

population and now are mainly confined to use on the coast of British Columbia. These

trucks are no longer manufactured and most are at least 30 years old. They have been

kept in service because of their robust construction and high load carrying capacity.

These log trucks commonly consist of a tandem drive axle tractor and a tandem axle pole

trailer.

Figure 1.1: Typical off-highway log truck

The typical loaded mass of these trucks is between 107 to 122 tonnes, with

payloads of approximately 68 to 83 tonnes (Oakley and Marshall, 1989). These trucks

were designed for off-highway operation where axle loads and vehicle dimensions are not

subject to regulations applied to public roads. Figure 1.2 presents maximum axle loads

for an off-highway log truck used for the design of forestry bridges in British Columbia.

1
In addition to the differences in size and weight between highway and off-highway log

trucks; there are also structural differences. Two of most common off-highway truck

models are considered in this study—the Hayes H D X and the Pacific P I 6 .

Steering axle Drive axle group Trailer axle group

kN 105 602 602

lb 23545 135320 135320

Figure 1.2: Off-highway log truck axle weights used in forest road bridge design (British
Columbia Ministry of Forests 1999)

Because of the heavy payloads drivers of these off-highway log trucks may have

difficulty braking on steep hills. Parker (2004) states that road grade, speed and mass of

the truck must be carefully managed when descending steep grades so that the required

retardation power does not cause excessive brake temperatures resulting in brake fade.

The Workers Compensation Board of British Columbia ( W C B , 2004) reported a fatal

accident for a truck driver descending a steep grade with a heavily loaded off-highway

truck. The recommendations in the W C B report suggested loads should be reduced for

steeper grades to ensure vehicle control can be maintained.

Off-highway truck payloads are difficult to assess because these trucks are not

equipped with on-board weighing systems and variations in wood density and load

2
dimensions make visual estimates highly inaccurate. To-date on-board weighing systems

have not been developed for off-highway log trucks because it is common to load them

until their volumetric capacity is reached rather than restricting loads to some maximum

allowable axle weight.

In 2004, the Workers Compensation Board of British Columbia asked the Forest

Engineering Research Institute of Canada ( F E R I C ) to develop a guideline for predicting

the safe maximum grade for descending with various off-highway truck payloads. To

utilise the guideline w i l l require trucks to limit their load size by measuring their axle

loads during loading. While off-highway pole trailers can be equipped with

.commercially available non-load-bearing transducers (Figure 1.3) the structural

differences between on- and off-highway tractor suspensions w i l l prevent the use of

highway tractor scales with off-highway tractors.

3
 Figure 1.3: Walking beam load transducer (Phillips 1990)

1.2 Background review

Numerous on-board load-measuring systems are available for highway log trucks.

These systems typically employ strain gauge technology or air pressure gauges mounted

on the air suspension. D u f f (2003) lists some common load measuring systems,

appropriate for use with highway log trucks. They include:

1) mounting the log bunk support pedestal and bunk roller ring on beams

instrumented as load cells (see Figures 1.4 and 1.5),

2) mounting displacement transducers between the frame and an axle,

3) plumbing air pressure gauges in the air bag suspension system,

4) instrumenting the trunnion shaft to act as a load transducer in a suspension

4
 having a single main trunnion shaft, and

5) mounting beam load cells between the trailer frame and spring hangers

(see Figure 1.6).

Figure 1.4: L o g bunk assembly on a highway Figure 1.5: Beam load cell mounted on a
log truck highway log truck (Phillips 1989)

The differences between off-highway and highway log trucks make it challenging

to adapt highway type load-measuring systems for use on off-highway tractors.

Specifically, when considering the load measuring systems listed above, the following

can be noted:

1) the bunk roller ring diameter is much larger in the off-highway tractor and it is

an integral part o f the tractor frame as seen in Figure 1.7. Therefore, it is not possible to

mount the bunk pedestal and bunk roller ring on beams instrumented as load cells.

5
Figure 1.6: Trailer suspension assembly Figure 1.7: Bunk assembly on an off-
with SI beam load cell (adapted from highway log truck
www. sitechnologies. com)

2) Hayes H D X tractors have a leaf spring suspension, similar to as shown in

Figure 1.8. The change in the clearance between a tractor frame and its drive

axles is a nonlinear function of the payload carried by the leaf spring

suspension and this would complicate calibration of a displacement transducer

system. In addition, individual leaf springs may crack without the driver

noticing and this could cause the displacement transducers to generate

erroneous load readings.

6
Figure 1.8: Leaf spring suspension of an off-highway tractor (adapted from Lathan 1999)

3) Heavy duty off-highway tractors generally do not use air bag suspension

systems.

4) Hayes H D X tractors use two short (140mm X 610mm) trunnion shafts. These

short trunnion shafts are completely encased in their trunnion saddles. G i v e n

the dimensions and confinement of the trunnion shafts, it is expected that the

strains generated in these shafts due to bending or shear w i l l be small.

Therefore, load measurements with the trunnion shafts would lack sufficient

variation to achieve adequate resolution. Further, construction and

instrumentation of specially designed trunnion shafts would be expensive, and

replacement of damaged shaft gauges would be a complicated and lengthy

repair.

7
5) For the off-highway tractors considered in, this study, the ends of the leaf

springs are supported by metal pads mounted in brackets that are attached to

the drive axle tubes. It might be possible to mount these brackets on shear

load cells and then mount the load cells on the drive axle tubes as shown in

Figure 1.9. Unfortunately, such an arrangement would require removal of the

drive axles to fit the load cells between the brackets and the axle tube. A l s o ,

because of the heavy payloads, the load cells would have to be relatively

thick. When installed, these load cells would raise the center of gravity of the

tractor by an amount equal to their thickness and thereby reduce the tractor's

roll stability.

Finally, the axle tubes could be used as load transducer by applying strain

gauges directly to the surface of the drive axle tubes. If strain gauges were to be

placed directly on the top surface of the drive axle tubes they might be well

protected, however, the stiffness of the robustly designed axle tubes may limit the

range of strains to be measured, and also any differences of inflation pressure

between the dual tires may affect the gauges' calibration (Newton 1987).

8
 Frame

( \{ ^ Leaf spring end support, t w

Figure 1.9: Load cells mounted between a drive axle and its leaf spring suspension

A review of the suspension components of a typical off-highway tractor revealed

that it might be feasible to instrument its suspension to predict axle weights based on

mechanical strain. The leaf springs and the U-bolts that fasten the leaf spring packs to the

trunnion shaft saddles both offered candidate locations for mounting strain gauges. In

order to be a useful indicator of payload the suspension component's strain should vary

linearly with payload, vary in a repeatable manner, and vary sufficiently over the range of

axle loads to provide adequate resolution. The main advantage of having a repeatable

linear calibration curve is that this greatly simplifies calibration of the system. This could

be an important consideration when off-highway log trucks are working in remote

locations without ready access to technical support or weigh scale stations.

O f the two candidate suspension locations, the U-bolt was preferred as a strain

gauge location over the leaf spring because the latter is believed to have non-linear strain

variation with load that may vary between successive load cycles (Newton 1987). In

9
addition, ease of replacement of damaged strain gauges might prove important given the

severe operating environment experienced by off-highway truck suspensions. However,

suspension U-bolts are normally pre-stressed, and this creates high initial strains that

could mask the incremental axial strains due to log loads. In order to measure a strong

signal on the U-bolt, it may be important to locate the strain gauges where a combination

of axial and bending strains occurs.

1.3 Objectives

The objectives of this research were to:

a) collect preliminary strain data from the suspension components of an off-

highway log tractor; and,

b) use a Finite Element M o d e l ( F E M ) of the trunnion saddle, U-bolt, and leaf

spring system of the off-highway log tractor to determine the locations on the U-bolt

surface that would provide the largest incremental strains when the truck is loaded. This

objective w i l l require the following:

i. Calibrate the F E M using the preliminary load and strain data;

ii. M o d e l the effect of preload on the incremental strain due to external loads;

and,

iii. Determine whether there is bending strain in the U-bolt that can be used to

increase the measured incremental strain.

10
1.4 Organisation of the thesis

The research has been carried out in two stages, namely preliminary strain

measurements and finite element modelling. Chapter 1 discusses the need for research

and the difficulty of retrofitting commercially available on-board axle weighing system to

an off-highway tractor. Chapter 2 discusses the preliminary testing and resulting

measurements from strain gauges placed on a leaf spring and a U-bolt of an off-highway

tractor suspension in June 2004. Chapter 3 describes the F E M of the suspension built to

evaluate strains along the entire U-bolt surface. Chapter 4 describes the effect of preload

and bending strains in the U-bolt as well as the F E M ' s calibration using the preliminary

test measurements. This chapter also describes how the calibrated F E M was evaluated for

potentially suitable strain gauge mounting locations. Chapter 5 draws conclusions from

this work and makes recommendations for improving the resolution of strain

measurements on U-bolts.

Appendix I describes the transformation of the strains along the curved portion of the

U-bolt from global to local coordinates. Appendix II includes the dimensions of the slip-

on tank used for preliminary strain measurement.

11
 CHAPTER 2

PRELIMINARY STRAIN MEASUREMENTS

2.1 Sensor location on the H D X suspension

In order to estimate wheel loads on a Hayes H D X heavy-duty off-highway log truck

consideration was given to instrumenting load transducers on all of the truck's axles.

Due to the front log bunk being located over the centre of the drive axles, however, the

steering axle load is believed to be insensitive to payload changes and therefore may not

need to be monitored. The pole trailer w i l l need to be instrumented and it can be

equipped with commercially available non-load-bearing transducers of the type used for

highway trucks (discussed in Chapter 1). The U-bolts that support the trunnion shafts on

the leaf springs, and the bottom leaf spring were selected as mounting locations for load

transducers on the drive axle suspension. It was not possible to instrument a trunnion

shaft during these initial strain tests because they were fully encased in the trunnion

saddles.

The installation of strain gauges to the. Hayes H D X tractor and subsequent load

testing was conducted in June 2004 at Hayes' service facility in Port Alberni, British

Columbia. One of eight U-bolt ends (four U-bolts) in the tractor suspension was

instrumented. Two gauges were installed on the U-bolt and aligned to measure strain

along the axis of the U-bolt (Figure 2.1). These strain gauges were located on the shank

portion of the U-bolt in order to avoid measurement variation caused by contact

interaction between the U-bolt and the spring pad. A third gauge was installed

longitudinally near the centre of the underside of the bottom-most leaf spring. This

12
mounting location offered the best combination of ample mounting space and (expected)

large bending strains, and it was far from any leaf spring stress concentrations.

U-bolt

Figure 2.1: Strain gauge locations

2.2 Instruments used

The strain gauges used were type C E A - 1 3 - 240UZ-120, (uniaxial gauges) from

Measurement Group, Inc. The stated gauge factor was 2.12 ± 0.5% at 24° C. Each strain

gauge was connected to a channel box (an SB-10 Switch and Balance Unit) as a single

active arm of a Wheatstone bridge (Figure 2.1). A Measurement Group System P3500

strain indicator was used to measure and record the strain values with a resolution of 1

microstrain. Four portable pad scales (model PT300) were used to measure the weight of

each drive wheel assembly in response to different payloads.

13
2.2 Methodology

The procedure for measuring tractor payload and load transducer output was as follows:

1. a slip-on water tank was installed on the log bunk of the Hayes H D X

tractor (Figure 2.2). The dimensions of the slip-on tank are given in

Appendix II;

2. strains were measured in the drive axle suspension members while the

volume of water in the slip-on tank was varied from full to empty.

The weight of water carried by the tractor at any time was calculated

as the estimated volume of water in the tank multiplied by a density of

1000 kg per m ; and,

3. the drive axle wheel loads were measured with the tractor parked on

four portable pad scales on a level paved surface.

Incremental strains and the wheel loads were measured in response to different

payloads carried by the tractor on a flat surface; no test loads were measured on a

slope.

14
 Figure 2.2: Measuring drive axle weights with portable pad scales

2.3 Results and Discussion

The weight of water in the slip-on tank and the resulting weights for the drive axle group

and for the right drive dual assembly are shown in Table 2-1. The maximum weight of

water added to the tractor corresponded to approximately 2 5 % of a typical full drive axle

group payload (i.e., 37 tonnes payload for a 55 tonne gross weight on the drive axle

group) for a Hayes H D X . The incremental differences show a strong correlation between

the weight of water carried by the truck and the total drive axle group weight. For

example, a total of 90 k N of water were removed from the tank and the resulting decrease

in total drive axle group weight was 87 k N . This indicates that additional payload added

to the truck would be carried almost entirely by the drive wheels and little or none would

be transferred to the steer wheels when the truck is on level ground. For travel on slopes,

15
the load shift onto or off of the steering axle could be estimated using a simple geometric

relationship.

Table 2-1: Estimated weight of water added and resulting drive axle group weights
Weight of Total drive axle Total right hand side
water group weight drive axle group weight
(kN) (kN) (kN)
121 279 144
110 270 140
101 261 135
93 254 131
85 246 127
78 238 123
69 230 119
61 223 115
54 214 111
48 207 107
37 198 103
31 192 99

Figure 2.3 illustrates the incremental strain response i n the leaf spring due to

unloading a total o f 90 k N o f incremental load from the tractor; and, the negative

incremental strains denote a relaxing of longitudinal strain as the load is decreased. The

leaf spring packs are subject to bending moments when loaded and unloaded, and this

bending generates relatively large strain responses (e.g., a 70 microstrain increase with a

90 k N load increment). The incremental strain in the bottom leaf was found to be linear

over most o f the range o f incremental load evaluated. However, leaf spring deflection

under a given load is known to be greatly affected by friction and may give variable

results (Newton 1987). A l s o , multi-leaf spring suspensions have different spring rates at

different load ranges, and the spring rate for intermediate loads is typically variable ( S A E

1999). Another concern with locating load transducers on the leaf springs is that the

leaves are susceptible to cracking and this would effect transducer calibration. Finally,

16
locating load transducers on the bottom o f leaf springs exposes the instrumentation to

road salt; aggregate and other debris thrown up by the tires while driving. For these

reasons leaf springs are not preferred as an axle weighing system.

Incremental load carried by the tractor (kN)

90 80 70 60 50 40 30 20. 10

Figure 2.3: Incremental leaf spring strain response to unloading the tractor with strain
gauges applied when tractor was loaded

Figure 2.4 presents the measured strains on the shank portion o f a leaf spring

U-bolt i n response to different incremental loads. The load resolution was approximately

± 18 k N (1.8 tonnes) for the range o f incremental loads tested. If this resolution was

consistent over the full range o f loading, the measurement error for a fully loaded drive

axle group would be 3.0% (i.e., 18 k N / 602 kN).

17
 Incremental load carried by the tractor (microstrain)
100 80 60 40 20

A Lower gauge location 1

• Upper gauge location 2 FT = 0 . 9 0 3 7

Figure 2.4: Incremental axial strain measured i n two locations on a leaf spring U-bolt
during unloading with strain gauges applied when tractor was loaded.

The incremental strains from both strain gauges had very similar slopes for the

range o f incremental loads measured. The strain at location 2 on the U-bolt also was

consistently about 1 microstrain higher than at location 1 for the range o f incremental

loads measured. Given the limited amount o f testing it cannot be conclusively stated

whether the difference i n measured incremental strains at locations 1 and 2 was due to

differences i n strain instrumentation or whether the leaf spring U-bolts are subjected to

bending.

In Figure 2.5 drive axle group weight was plotted against measured incremental

strain. Each level o f incremental strain corresponded to two or three different drive axle

group weights. These weights were averaged, therefore, and reported as a single average

18
value versus that level of incremental strain. The result shows the linear relationship with

drive axle group weight and strain, however, the magnitude of the strain variation with

payload was small (i.e., no more than 5 microstrains up to a maximum payload of 90 k N ) .

U-bolts in the Hayes H D X suspension are normally tightened (preloaded) in order to

ensure the leaf springs remain clamped, and therefore under high axial tension even when

the truck carries no payload. The preload was not known and was indirectly estimated

from the manufacturer's torque specification. The preload in the U-bolt could be

estimated using the following formula that relates torque and bolt cross-sectional area

(diameter) to preload (Dayton 2001):

Preload = - I ^ H _ e
( 1 )

K x Diameter

The torque coefficient, K, is a measure of the friction between the nut and the U -

bolt threads. A value of 0.2 was specified for K , as per conventional practise which

assumes that the bolt is new and lubricated (Dayton 2001). The torque for the leaf spring

U-bolts was taken to be 1760 N - m , based on discussions with Hayes Forest Service about

their torqueing practice. Using the above parameters, formula (1) estimates a preload of

231 k N for a 36.75 mm-diameter U-bolt. The small strain variation for the range of

payloads tested may have been due to the large amount of preload in the U-bolt (Norton

1996) and because the maximum test payload was only about 2 5 % of a full payload.

However, the linear relation in Figure 2.5 indicates that U-bolts have some potential as

load transducers provided that their strain rate is repeatable.

19
 310

170 --

150 -I : 1 , , 1 1 1 , , 1

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5

Incremental strain (microstrain)

Figure 2.5: Incremental axial strain at location 1 on the suspension U-bolt i n response to
an averaged drive axle group weight.

Given the findings from the preliminary measurements, the following can be said

about measuring drive axle group loading with U-bolt strain:

1. The entire U-bolt surface should be analysed to determine where incremental

strains are largest and therefore would provide the greatest strain variation;

2. Strain gauging locations should be selected to avoid contact interaction effects

that could prevent measurement repeatability; and,

3. Satisfactory weight measurements may be possible despite the limited strain

variation observed in the U-bolt i f a higher resolution strain indicator and

higher resistance strain gauges are used.

20
 CHAPTER 3

FINITE E L E M E N T M O D E L C O N S T R U C T I O N

3.1 Overview

The preliminary measurements (Chapter 2) indicated that the axial strain developed in the

leaf spring U-bolts varied linearly with increasing payload. A n F E M was created to

check strain distribution over the entire U-bolt surface in order to optimise the placement

conditions of strain gauges. Due to their design U-bolts can be subject to a bending load.

The F E M w i l l allow an examination of the U-bolt surface for areas subject to increased

strain due to a combination of axial and bending strains.

Two aspects of the U-bolt problem complicate the analysis. First the interaction

between the U-bolt and leaf spring is a contact problem. Second, the U-bolt is under a

significant preload. Thus, the U-bolt does not sustain the full effect of the applied load

(Norton 1996). Both of these complicating aspects can be modeled using ANSYS®. The

procedure for a typical structural analysis is to define the system's physical components,

transform it into a structural model with meshing (i.e. descretization), apply appropriate

load and boundary conditions until model predictions are comparable to measured strain

responses, and finally interpret the F E M results. Figure 3.1 shows a framework used for

modelling the U-bolt in this work. Section 3.3 describes the geometric model and F E M .

Section 3.4 describes the boundary condition and modelling methodology. Post-

processing results are discussed in Chapter 4.

21
 Define geometric model
(geometry o f I -boh. leaf
spring, and trunnion saddle)
O
CO

o
C
i
D i s c r e t i z e the m o d e l (meshing
OH
the m o d e l )

D e f i n e the b o u n d a r y c o n d i t i o n s
(contact a n d displacement
boundary conditions)

tij
03
to
C
o L o a d s t o p 1: p r e l o a d t h e I. -bolt
( a p p l y t e n s i o n w i t h a t h e r m a l >lrain)
o
r

Load step 2: apply an external load

(equal to the payload applied in testing)
V )

Calculate strain distribution and

Post-processor interpret. Revise meshing, boundary
static conditions and preloading until outputs
agree with measured test strains.

Figure 3.1: F E modelling procedure for the U-bolt analysis

3.2 Background of U-bolt modelling

Considerable research has been devoted to analytical, mechanistic and numerical models

simulating connections using standard bolts (i.e., straight, threaded bolts with a head).

22
V e r y little research on U-bolts, however, is documented in the literature. T w o studies

involving U-bolts are discussed below.

Diamantoudis and Apostolopoulas (2002) developed a three dimensional (3-D)

model to determine whether the stress in two U-bolts, that secured a plate to a truck

frame member, exceeded a limiting stress. The steel plate and the frame member of the

truck were modeled with 3-D solid elements. The U-bolts, however, were modeled more

simply with 2-D beam elements (the nodes have three degrees of freedom: translation

along the X and Y axes and Z-axis rotation). Figure 3.2 illustrates the U-bolted plate and

frame member obliquely and in cross-section. They found that the bending stresses in the

C-channel were highest at its bottom corner where the influence of the eccentric vertical

loading from the U-bolt shank was greatest. Based on this an alternative mounting

connection was recommended.

_ —

U-bolt

Figure 3.2: Finite element modelling of U-bolted assembly (Diamantoudis 2002)

K i r b y and Charniga (2005) modeled U-bolts clamping a leaf spring pack, axle

seat and U-bolt lower bracket to an axle tube. This analysis was to investigate twisting of

the leaf spring on the axle seat in order to address a noise concern with the suspension

23
system. The U-bolt was used only to simulate the structural response o f the leaf spring to

preload. The U-bolt shanks merely transferred preload to the curved portion and, i n

order to reduce computational time, were meshed with 2 - D beam elements (Figure 3.3).

U s i n g this model the simulated load at critical points on the axle seat was correlated with

measured loads for different coefficients of friction. Friction coefficients i n the range o f

0.1 to 0.15 were simulated.

Curved portion of U-bolt

Figure 3.3: F E M of a U-bolted leaf spring assembly (Kirby 2005)

It can be inferred from this review that little work has specifically concentrated on

analysing incremental strain in the U-bolt. The objective o f the thesis is to analyse the

maximum incremental strain along a U-bolt i n order to maximise strain resolution.

Therefore, in this thesis the U-bolt was modelled completely with 3-D solid elements.

24
3.3 F E M description

The F E M begins with the construction of a 3-D geometric model of the U-bolt, leaf

spring pack and trunnion saddle assembly based on physical measurements of

components (Figure 3.4). Friction between the leaf spring leafs and between the leaf and

the spring pad was assumed to be high enough to restrict differential motion between

them in the region of the U-bolt. Given this simplification and that our interest was

confined to the U-bolt and its zone of contact with the spring pad and leaf springs

(hereafter called the leaf spring) the leaf spring was modelled as a single block. A similar

assumption was made in the region of the U-bolt ends and nuts and this allowed them to

be modelled as i f they were glued together and to the bottom of the trunnion saddle. In

addition, the trunnion saddle had a complex geometry, however," it was modelled as a

block to reduce its modelling requirements. The overall number of nodes available for

modelling is limited by A N S Y S and these simplifications resulted in more nodes being

available for modelling the U-bolt. More nodes were required for modelling the U-bolt

because an accurate strain distribution was needed for comparison with the field

measurements and for investigation of the U-bolt surface to identify suitable gauging

locations.

25
 152.4mm/ 6inch Leaf spring block

4-
36.75 mm /
1.5 inch

27.56 mm/l.125 inch

Nut ?
T r u n n i o n Saddle 110.25 m m / 4 . 5 i n c h

Leaf spring block length (L)

Figure 3.4: Schematic of the U-bolt assembly

Next a structural model was created with S O L I D 9 2 elements sourced from the

A N S Y S element library. These quadratic 3-D tetrahedral elements were used instead o f

simpler 3-D linear tetrahedral elements because the S O L I D 9 2 elements are more suitable

for estimating strains i n curved sections and are generally more accurate ( A N S Y S 2004).

The S O L I D 9 2 element has 10 nodes with three degrees o f freedom at each node (Figure

3.5).

26
 Node

Figure 3.5: 3-D 10-node tetrahedral structural solid

In order to ensure compatibility of the mesh elements only one type of element

(i.e. S O L I D 9 2 ) was used for modelling all components of the assembly.

When loaded, the U-bolt assembly undergoes a relative displacement at the contact

interface (Figure 3.6) and this must be accounted for in order to obtain an accurate

predication of strain along the surface of the U-bolt. To model the relative displacement,

surface-to-surface contact elements were placed between the curved portion of the U-bolt

and leaf spring block, and also between the leaf spring block and the trunnion saddle

(Figure 3.6).

Surface-to-surface contact elements were modelled using A N S Y S contact 174

and target 170 elements. These types of elements are capable of transferring forces and

stiffness between the surfaces. The contact elements take the shape of the underlying

elements (10-node tetrahedron in this F E M ) and therefore appeared as triangular-shaped

elements (Figure 3.7).

27
 Contact elements

Contact elements

Figure 3.6: Location of contact elements

Surface contact element

Figure 3.7: Surface-to-surface contact elements

28
The material used in the F E M was assumed to be linear elastic. The physical properties

of individual suspension components are listed in Table 3-1.

Table 3-1: F E M component physical properties

Component Type of steel Property Modulus o f Poisson's Thermal expansion

direction elasticity ratio coefficient at 21 °C
GPa 10" m/m/°C
6

U-bolt 4140 (Alloy) Isotropic 210 a

0.291 a
12 b

Leaf spring 5160 (Alloy) Isotropic 200 c

0.300° 13.5°
Trunnion Cast Isotropic 200 c
0.300° 13.5°
saddle
a
Walsh 2000
b
Speck 1997
° A S M 1994

3.4 Meshing of the F E M

The meshing algorithm i n A N S Y S allows the degree o f mesh coarseness to be

selected by the analyst from a scale of 1 to 10, with 10 being coarsest. The entire U-bolt

was meshed moderately densely (i.e., with a mesh coarseness o f 6). A denser mesh (i.e.,

with a line refinement level o f 2) was applied around the outside surface o f the curved

portion o f the U-bolt and around the inside surface where the U-bolt contacted the leaf

spring. A finer mesh results in the boundary of the body in the model being closer to the

actual shape o f the body and this reduces the effect of stress concentrations due to

meshing (Saravi and Lyons, 2004). The rest of the F E M components (i.e., leaf spring and

trunnion saddle) were meshed coarsely (i.e., with a mesh coarseness o f 8) because the

29
overall number of nodes available was limited. Figure 3.8 illustrates the F E M o f the

U-bolt assembly.

Figure 3.8: F E M of the U-bolt assembly

3.5 Boundary conditions

The lower edges of the leaf spring were constrained to prevent translation in the X , Y and

Z direction (Figure 3.9).

30
 Leaf spnng
edges
constrained
inX,Y,Z

Figure 3.9: Boundary condition

3.6 Preload modelling

In the first loading step, the U-bolt was preloaded. The preload i n the U-bolt can be

modeled using thermal strain or built-in A N S Y S P R E T S 1 7 9 elements (Imaoko 2000).

Stalling (1992) simulated preload i n a bolt with thermal strain created by lowering the

temperature of the components. The bolt was shrunk until the dimension changes created

the desired level of tension in the bolt. In the analysis of the H D X truck suspension i n

this thesis, the U-bolt preload was simulated using thermal strain. The basic procedure

for modelling preload in the U-bolt was:

1) assign an appropriate coefficient o f thermal expansion to the suspension

components,

31
 2) specify a uniform temperature (21° C) to the components,

3) specify a subzero temperature for the shanks of the U-bolt (Figure A.I),

4) compute the U-bolt axial tension and,

5) iterate until the desired preload is achieved.

3.7 External load

In the second loading step, an external load was applied to the bottom part of the trunnion

saddle as a uniformly distributed pressure (Figure 3.9).

3.8 Contact modelling

Contact element simulate contact between two surfaces by generating contact forces of

finite stiffness value when two surfaces approach each other and by removing the contact

forces with stiffness as zero when the surfaces move away (Rizzo 1991). It also

simulates sliding with friction. There are two common algorithms to simulate this

contact element:

1) Penalty method. With this method, springs are placed in the contact interface.

Some finite amount of penetration is required mathematically to maintain

equilibrium; however, physical contacting bodies do not interpenetrate. Increasing

the stiffness of the element reduces the magnitude of the penetration, however, the

condition of the stiffness matrix depends on the contact stiffness itself. If the

contact stiffness is too large, it w i l l cause convergence difficulties.

2) Lagrange multiplier method. This method adds the extra degree of freedom to

the stiffness matrix to satisfy the constraint exactly. In this method no imaginary

32
 penetration is assumed, therefore error free solution is granted for sticking friction

(Zahavi and Barlam 2001).

The accurate strain distribution along the curved surface of the U-bolt was

important in this analysis. Therefore, a "Lagrange multiplier on contact normal and

penalty on tangent" algorithm ( A N S Y S 2004) was used for modelling the contact

interaction between the curved surfaces of the U-bolt and the .leaf spring. This option

uses a Lagrange multiplier on the contact plane and enforces zero penetration while a

displacement penalty on the tangent plane limits slip for a sticking condition ( A N S Y S ,

2004). A default algorithm, "Augmented Lagrange", was used for modelling the contact

between the leaf spring and the trunnion saddle in order to save computational time.

A N S Y S solves for displacement at contact surfaces by adding the stiffness matrix

of the contact surface to the stiffness matrix of the bodies under consideration so that, for

any load increment, equilibrium is achieved. The equation that links force, displacement

and stiffness and has the following form:

[K + K ]u=f
b c (2)

where Kb is an n x n stiffness matrix of the bodies under consideration,

K is an n x n stiffness matrix of the contact surface,

c

u and f are the displacement and the load vector respectively having an n x 1

form (Zahavi and Barlam 2001).

When contact friction is introduced in an F E M the resulting slip can generate an

unsymmetrical stiffness matrix (Burke and Olatunbosun, 1997). A t higher coefficients

of friction, when slip is significant, a solution may not converge (that is, may not be

possible). In such cases, A N S Y S recommends using an unsymmetrical stiffness matrix

33
to improve the likelihood of convergence. Solving with an unsymmetric matrix,

however, is more computationally expensive than solving with a symmetric matrix. A

symmetric matrix was used to solve equation (2) because of limited computing memory

and because slip was assumed not to be significant in the region of U-bolt to leaf spring

contact. Laursen and Simo (1994) developed a symmetrization algorithm by which most

frictional contact problems can be solved using symmetric stiffness matrices. ANSYS

features this symmetrization algorithm and this was used for the analysis.

3.9 Post processing

The longitudinal strain was calculated along the U-bolt surface using the PATH

command ( A N S Y S 2004). This command interpolates stress, strain and displacement

results between adjacent nodes along a straight line between two specified end points. In

order to estimate the strain distribution along the U-bolt surface the straight, shank

portion was analysed separately from the curved portion. The strains along the U-bolt

shank were expressed in conventional global coordinates: x, y, z (Appendix I). To

simulate strains measurable on the curved surface, however, strains in the curved portion

of the U-bolt were expressed in local coordinates (i.e., oriented in longitudinal, transverse

and normal directions to the surface (Appendix I)). A macro was created to transform the

strains along the curved portion of the U-bolt from global to local coordinates, and this is

included in Appendix I. Following these calculations, the incremental strain due to the

external load applied to the U-bolt was calculated as the difference between the strains

obtained in load step 2 (i.e., preload and incremental load) and load step 1 (i.e., preload

only).

34
 CHAPTER 4

FINITE E L E M E N T M O D E L L I N G R E S U L T S

4.1 Overview

Section 4.2 describes the sensitivity analyses that were performed to determine what leaf

spring length was necessary to minimise the effect of the boundary conditions on the

incremental strain along the U-bolt surface, and to determine the appropriate magnitude

of the preload and the coefficient of friction. Section 4.3 presents an evaluation of

incremental strain along the U-bolt under the loading conditions encountered in the

preliminary field measurements. Based on the results recommendations were made for

locating strain gauges on U-bolts for an axle weighing system. Section 4.4 discusses

possible sources of bending in the U-bolt shank. Section 4.5 discusses the limitations of

the empirical measurements and finite element modelling.

4.2 Sensitivity Analysis

Although many variables affect U-bolt behaviour under load this analysis specifically

investigated the effects of preloading and friction between the U-bolt and leaf spring

block. A range of typical friction coefficients (from 0.1 to 0.3) was considered because

the surface roughness was not known. A range of preload value from 142 k N to 303 k N

was selected with the estimated preload from section 2.3 being mid range. A n analysis

was also undertaken to find the minimum length of the leaf spring for which the

incremental U-bolt strains did not change. The length of the leaf spring block was varied

in order to reduce its modeling requirements and, at the same time, minimising boundary

35
condition effects. The mesh properties of the model, described in Section 3.4, were kept

constant throughout the sensitivity analysis.

4.2.1 Leaf spring block length

A leaf spring pack on a Hayes H D X is 1778 m m long. The leaf spring pack was modeled

as a solid block (with the leaf springs glued to one another) and also was shortened in

order to reduce the number of nodes required to model it. The boundary condition used

at the bottom edge of the leaf spring (indicated in Figure 3.9) often creates stress

concentrations ( A N S Y S 2002). It was expected that these stress concentrations would

reduce the accuracy of strain estimates on the U-bolt, and that shortening the leaf spring

block would exacerbate this effect. Therefore, a sequence of analysis runs was

conducted in which the length of the leaf spring was increased until this boundary

condition effect began to decrease. The parameters used in the analysis of leaf spring

block length, and their values, are listed in Table 4-1. A preload value of 142 k N

(corresponding to a torque of 1085 N-m) was used for this analysis. L e a f spring block

length was varied between 165 and 521 mm. For block lengths less than 165 mm, the

boundary condition effect adversely affected the strain estimates on the U-bolt. For leaf

spring lengths greater than 521 m m the number of nodes required for modeling became a

limitation. The uniform load applied to the bottom of the trunnion saddle was equivalent

to 22.5 k N (i.e. the average difference between the right hand side drive wheel weights

when fully loaded with water and when unloaded, refer to Table 2-1). Friction

coefficient was taken to be 0.2 (i.e., a mid-point in the range of typical values (0.1 to 0.3).

36
Table 4-1: R u n sequence inputs to evaluate leaf spring block length
Run Preload Incremental Coefficient Leaf spring length*
sequence (kN) load (kN) of friction (mm)
1 142 22.5 0.2 165, 216, 267, 368, 419, 470, 521
1 . 1 1 1 : _£

Figure 3.0.4 describes the physical dimensions of the leaf spring suspension

Figure 4.1 illustrates the relation between leaf spring block length and incremental

strain at gauge location 1 on the U-bolt. It can be seen that variation in the incremental

strain with respect to increasing spring block length is reduced for leaf spring block

lengths of over 400 mm. The same finding was true for incremental strains at gauge

location 2 (Figure 4.2). Therefore, a leaf spring block length of 470 m m was used in all

subsequent analyses.

100 200 300 400 500 600

Leaf spring block length (mm)

Figure 4.1: Incremental strain at gauge location 1 on the U-bolt for various leaf spring
block lengths (incremental load of 22.5kN, preload of 142 k N and U-bolt to leaf spring
coefficient of friction of 0.2)

37
 3.4

3.2 -

CD 3"
w
e
.9
£ 2.8 --
c
to
2 2.6 -
c
CO
E
CO

o 2.4 -

2.2 - :

2 --
0 100 200 300 400 500 600
Leaf spring block length (mm)
Figure 4.2: Incremental strain at gauge location 2 on the U-bolt for various leaf spring
block lengths (incremental load of 22.5kN, preload o f 142 k N , and U-bolt to leaf spring
coefficient of friction of 0.2)

4.2.2 Preload

A sensitivity analysis was conducted on U-bolt preload to determine the most appropriate

preload level for modelling. This analysis consisted o f evaluating preload at five levels

and with five different coefficients o f friction, making for a total o f 25 observations.

Table 4-2 summarises the input parameters used in this sensitivity analysis. The torque

specification for the leaf spring U-bolts was 1630 N - m , however, this torque can vary i n

practise by 270 N - m or more (Waugh 2005). Therefore, torque values between 1085 and

2310 N - m were considered i n the analysis. The preload was calculated using

equation (1) for the given torque range.

38
Table 4-2: R u n scenario for the different levels of preload and coefficient of friction
between the curved portion of U-bolt and leaf spring
Run Preload Incremental Leaf spring block Coefficient of friction
sequence (kN) load (kN) length (mm)
2 142 22.5 470 0.1,0.15,0.2, 0.25,0.3
3 196 22.5 470 0.1,0.15,0.2, 0.25,0.3,
4 231 22.5 470 0.1,0.15,0.2, 0.25,0.3
5 267 22.5 . 470 0.1,0.15,0.2,0.25,0.3
6 303 22.5 470 0.1,0.15,0.2, 0.25,0.3

Figures 4.3 and 4.4 illustrate how U-bolt incremental strain varied with preload

over the range of contact friction values, at an incremental load of 22.5kN. The

simulated strains were compared to the measured incremental strains for the estimated U -

bolt preload. The preload value for the strain-gauged U-bolt in the preliminary test was

estimated to be 231 k N (Section 2.3),

The results of this sensitivity analysis indicated that:

1. U-bolt incremental strain was insensitive to preloads of between 140 k N and 231

kN;

2. Incremental strain generally increased with preloads above 231 k N for all values

of coefficient of friction;

3. Variation in incremental strain due to the coefficient of friction between the

U-bolt and the leaf spring increased for preloads above 231 k N ; and,

4. Variation in incremental strain with the coefficient of friction between the U-bolt

and leaf spring was greater at gauge location 2 than at gauge location 1.

39
 • preliminary strain m e a s u r e m e n t

3 . 0 -I , , , 1 , , , ,—

140 160 180 200 220 240 260 280 300

P r e l o a d (kN)

Figure 4.3: Incremental strain at gauge location 1 on the U-bolt for various preloads and
coefficients of friction (u) between the U-bolt and the leaf spring (incremental load of
22.5 k N and leaf spring length of 470 mm)

5.5 T

5.0

_ 4.5

.£ 4.0

3.5

3.0
-A-u-0.2

-e-u-0.25
2.5
-*-u-0.3

• preliminary strain m e a s u r e m e n t
2.0
140 160 180 200 220 240 260 280 300 320

P r e l o a d (kN)

Figure 4.4: Incremental strain at gauge location 2 on the U-bolt for various preloads and
coefficients of friction (u.) between the U-bolt and the leaf spring (incremental load of
22.5 k N and leaf spring length of 470 mm)

40
 Although larger incremental strains were calculated for preload values in excess

of 231 k N these may not be obtained, in practise, because of the relatively high level of

preload required to generate them. The U-bolt preload from field testing was estimated

to be 231 k N and this value was taken for all subsequent F E M analyses.

4.2.3 Coefficient of friction

The parameters given for run sequence 4 in Table 4.2 were used to estimate a suitable

coefficient of friction between the U-bolt and the leaf spring block. Figures 4.5 shows

the simulated incremental strain at locations 1 and 2 on the U-bolt for different

coefficients of friction. N o clear trend for the relation between incremental strain and

coefficient of friction was observed; different patterns are observed for different

estimated preloads in Figure 4.3 and Figure 4.4. Incremental strain at gauge location 1

was found to be relatively insensitive to coefficient of friction while incremental strain at

gauge location 2 was a maximum at a coefficient of friction of 0.2. Due to the generally

inconclusive results of the effect of coefficient of friction on the incremental strain a

midrange value of 0.2 was selected for all subsequent analyses.

41
 5.5 -i
o Simulated incremental strain at
location 1
• Simulated incremental strain at
location 2

c
CD
CO
P
•| 4.5
c
to

£ 4
CD
E
CD
O
c
3.5

0.05 0.1 0.15 0.2 0.25 0.3 0.35

Coefficient of friction

Figure 4.5: Simulated incremental strains at locations 1 and 2 on the U-bolt for different
U-bolt-to-leaf spring coefficients o f friction (incremental load o f 22.5 k N , preload o f 231
kN)

4.3 Analysis o f incremental strain

Based on the results from the sensitivity analysis, the distribution o f incremental strain

along the entire surface o f the U-bolt was calculated using a leaf spring block length o f

470 m m , an estimated preload o f 231 k N , external load o f 22.5 k N and a coefficient o f

friction o f 0.2. Figure 4.6 shows the distribution o f incremental strain along the U-bolt

surface. A s per the analysis objective, this distribution was examined to determine the

best possible locations for mounting strain gauges on the U-bolt surface.

42
 0 50 100 150 200 250 300 350 400
Distance along the U-bolt from the trunnion saddle (mm)

Figure 4.6: Distribution of incremental strain along the outer edge o f U-bolt surface
(incremental load of 22.5 k N , preload of 231 k N and friction coefficient o f 0.2 between
the U-bolt and the leaf spring)

Figure 4.6 indicates that the maximum incremental strain occurs close to the

trunnion saddle and that it decreases uniformly from 0 m m to 220 mm. Where the U-bolt

curved portion joins the straight, shank portion the incremental strain changes rapidly and

reaches a minimum (0.1 microstrain) at 285mm. Incremental strain is relatively constant

in the curved portion o f the U-bolt from 380 mm to 450 m m (i.e., within 70 m m o f either

side of the apex of the curve). To investigate the source of bending the incremental strain

on the inner surface o f the U-bolt was calculated and compared to that of the outer

surface. Figure 4.7 presents the incremental strain distributions along both inner and

outer surfaces o f the shank portion of the U-bolt. The increase in incremental strain on

the inner surface towards the curved portion of the U-bolt was likely generated by a

43
superposition o f both axial and bending strains. Therefore, where the strain curves

intersect the bending strain approaches zero as it changes sign and the incremental strain

at this point (3.21E-06 microstrains) is solely due to axial load. The incremental strain

due to bending can then be computed for any location on the shank by subtracting

3.21E-06 microstrains (the strain due to axial load) from the total incremental strain.

Figure 4.8 shows the resulting strain distribution due to bending along the outer surface

of the shank. The change in sign o f the incremental strain at 205 m m indicates that this is

an inflection point for the displacement.

» Outer surface of U-taolt

54
a Inner surface of U-bolt

O D

£
•
g outer

£ 2
Shank portion

0 50 100 150 200

Distance along the U-bolt shankfromthe trunnion saddle (mm)

Figure 4.7: Incremental strain distribution along the inner and outer surfaces o f the U-bolt
shank (incremental load 22.5kN, preload of 231 k N and friction coefficient o f 0.2)

44
 4

outer

-
S h a n k portion j
N

h
" • — • N \
1 ;

I
•
«•

•
•
•
*
•
•
•
*
•
•
•
•
*
*

(t 50 100 150 200 •

Distance along the U-bolt shank from the trunnion saddle (mm)

Figure 4.8: Bending strain distribution along the outer surface o f the U-bolt shank
(incremental load o f 22.5kN, preload o f 231 k N and friction coefficient o f 0.2)

The F E M predicted that the largest incremental strains on the shank o f the U-bolt

occur near the trunnion saddle (Figure 4.6), however, a more preferred location for the

strain gauge would appear to be within 70 m m to either side o f the U-bolt apex because:

1. incremental strain values are relatively constant within this area and so w i l l be

accurately read by strain gauges,

2. incremental strain values are due to axial and bending strains that when

combined generate close to the maximum incremental strain i n the U-bolt,

3. the gauge location is relatively far from stress concentrations created by the

fixed end conditions at the trunnion saddle and nut, and

4. the location is better protected from debris and spray than most other parts o f

the U-bolt.

45
 A s seen in the.field testing and in the F E M results, the incremental strain in the

U-bolt was very small. The ability of strain gauges to detect small changes in strain is

considered to be infinite, however, a resolution of 0.1 microstrain is the smallest practical

value attainable because of the limitation of instrumentation and other performance

factors (Window 1992). Acceptable weight measurement accuracy can be achieved with

a strong signal strength, however, even i f strains are relatively small. The signal output

of strain gauges can be improved by increasing the strain gauge sensitivity (Gauge

Factor) or adding a signal amplifier. Strain gauge sensitivity can be improved through

the use of higher resistance strain gauges supplied with higher input voltage. If an

amplifier is used it should be located close to the strain gauges so that a minimum of

noise from connecting wires is amplified.

In order to convert the small change in resistance of the strain gauge(s) into

voltage suitable for amplification and processing, a half bridge Wheatstone bridge can be

used (Window 1992). Figure 4.9 illustrates the recommended gauge location on the

U-bolt surface where the two uniaxial gauges are used in opposite arms of the bridge so

that the signal measured is from the combined axial and bending strains and two more

resistors are used to complete the half. Additionally a strong signal can be obtained

through series of similar half bridge system installed near the U-bolt apex.

46
 Figure 4.9: Recommended gauge location on the U-bolt

4.4 Possible causes of bending in the U-bolt shank

One possible source of the bending in the U-bolt shank observed in the F E M may be a

result of the curved portion of the U-bolt trying to retain its curvature while it is pulled in

the loading direction (Figure 4.10). In response, the shank portion of the U-bolt is bent

inwards at the intersection with the curved portion. This inward deformation would be

resisted by the fixed support at the bottom of the U-bolt shank. (The U-bolt end is

considered to be fixed because the shank portion of the U-bolt and nut are glued to the

trunnion saddle in the F E M ) . This deformation in the shank would create bending strain

(that is, tension on the outer surface and compression on the inner surface of the shank).

47
 curved
portion of
U-bolt

T- Tension
C- Compression
P- Axial force

Shank
portion of
U-bolt

/////

Figure 4.10: Bending in the U-bolt

A 2 - D model (Figure 4.11) was developed to check the above hypothesis. T w o

cases were evaluated: case 1 (Figure 4.12 Case 1) defines contact elements on both the

curved and straight portions; case 2 (Figure 4.12 Case 2) had contact only between the

curved portions o f the U-bolt and leaf spring block. In case 1 the U-bolt shank is

constrained against penetrating the leaf spring block; In case 2 the U-bolt shank is

allowed to penetrate the leaf spring block. Both cases were considered because, i n

practise, the shank portion o f the Hayes H D X suspension U-bolts is slightly wider than

the leaf spring pack and may or may not be i n contact along the straight portion o f the

48
U-bolt when preloaded. Figure 4.13 shows the response to axial loading for case 1 when

a 34.47 M P a (5000 psi) pressure was applied at the end of the curved portion of the

U-bolt. The contact with the leaf spring block prevents inwards deformation of both the

curved portion and the shank of the U-bolt. Thus, the shank deformation w i l l be purely

axial and no bending moments w i l l be generated in the shank portion of the U-bolt.

Figure 4.14 shows the deformed shape for case 2. The resulting deformation supports the

hypothesis that the curved portion of the U-bolt w i l l tend to retain its curvature under an

applied axial load.

14 2005
4 : IB : 59.

Figure 4.11: Geometry of two-dimensional model

49
Lines on the
contact surface
implies contact
surface's outwan
normal

>

1 ,

Case 1
Case 2
(a)
(b)

Figure 4.12: Contact elements in two cases

DISPLACEMENT
AN
NOV 14 2005

DMX =.00183

Figure 4.13: Deformed shape for case 1

50
 ANSYS
STE?=1 NOV 14 2005
SUB =5 14:18:23

Figure 4.14: Deformed shape for case 2

A second possible source of the bending in the shank could be the presence o f

misalignment in the model. In finite element modelling, the alignment of the geometric

model is confined to keypoints (Figure 4.15). In the process of model creation, the U-bolt

shank was created by extruding the cross-sectional area (Areal) to the prescribed height

(Figure 4.15) defined between keypoints 1 and 2. It was observed that these keypoints

were slightly misaligned in the z direction (by an order of magnitude 10" mm). This
15

small misalignment is not a significant source of bending moment for the U-bolt shank

FEM.

51
 curved portion of
U-bolt

keypoint 1 j
z

Area 1

|\shank
portion
of
U-bolt

keypoint 2

Figure 4.15: Defining U-bolt Volume in F E M

4.5 Limitations of the empirical measurements and finite element modelling

Empirical measurements were performed on the single truck parked on level ground. The

load applied was comparable to about one third o f the full payload. Incremental strain

was measured as load was removed from the truck and was measured i n only one

52
suspension U-bolt. The strain measurements may differ for trucks having different

suspension arrangements. Strains in the instrumented U-bolt would be expected to be

higher when the truck is on favourable grades because load shifts from the trailer to the

tractor, however, this load shift may easily be predicted through geometric relations.

Strains may also vary when load is added to the truck because of differences in the way

the suspension moves during loading versus unloading. The empirical data was not

checked for repeatability of incremental strains for payloads of similar magnitude.

Finally, measured strains may vary between the four U-bolts.

The F E M was correlated with only one set of load and incremental strain results

from the preliminary strain measurements. F E M inputs included assumed levels of U -

bolt preload and U-bolt-to-leaf spring coefficient of friction. A maximum U-bolt preload

was estimated using a preload-torque relationship but preload may vary, i n practise, for

different torque levels and torque coefficients. U-bolt-to-leaf spring friction may increase

with corrosion (in-service life), however, the F E M found no trend for coefficient of

friction so how this would influence strain measurements is unclear. A moderately fine

mesh was used to model the U-bolt however more accurate results are not expected to be

obtained with use of a finer mesh. This is because comparable results were obtained with

a finer mesh (not reported in the thesis) but required much longer computational time.

The U-bolt was modelled assuming that the straight shank portions were free to displace

inwards in response to bending moment. In practise, it is unclear whether the U-bolt

shanks could bend inwards and, i f they are prevented from this movement then the F E M

w i l l slightly overestimate strain levels in the U-bolt shank.

53
 CHAPTER 5

SUMMARY, CONCLUSION AND RECOMMENDATION

One way to reduce the incidences of brake failure of off-highway log trucks when

descending steep grades is to manage the trucks' load sizes. This can be done by

measuring the axle weights during loading. A review of on-board weighing systems i n

Section 1.2 reported that to-date no commercially available on-board technology exists

for measuring drive axle weights of off-highway tractors. This is because none of the

systems are made specifically for off-highway tractors and the on-highway truck systems

are not readily adaptable to the structural differences present in off-highway log trucks.

Due to the unsuitability of the existing technology, the suspension components of an off-

highway log truck were evaluated for use as load transducers. This evaluation was

conducted i n two stages: preliminary strain measurements were gathered from a loaded

off-highway tractor; and, a U-bolt from the tractor's leaf spring suspension was analysed

using finite element modelling. The preliminary strain measurements indicated that the

load-strain response of the U-bolt was linear for unloading over the tested range of

external loads. A three-dimensional F E M was created to check the outer surface of the

U-bolt for locations that developed the largest strain responses. The modelling was

carried out in two loading steps. First, the U-bolt was preloaded; and, second, an external

load was applied to the bottom of the trunnion saddle. A sensitivity analysis was

conducted to determine the appropriate level of preload, coefficient of friction, and leaf

spring block length for use in modelling the U-bolt incremental strain. The FEM

predicted that U-bolt incremental strain was a maximum near the trunnion saddle and a

54
minimum at the intersection of the shank and curved portions, and again near the

maximum at the top of the curved portion. The predictions of strain in the U-bolt shank

could be less reliable because they assumed no contact between the leaf spring block and

U-bolt shank. In practise, there may not be sufficient clearance to develop the bending

forces that generated the predicted strains. The incremental strains in the top of the

curved portion of the U-bolt were relatively constant and were close to the largest

observed in the U-bolt. Given the uniform strain distribution and the magnitude of the

incremental strains in the curved portion of the U-bolt it is the most promising location

for strain gauging.

The calibration of the F E M model constructed in this study was limited by the

lack of empirical data. However, the results of the incremental strain analysis indicate it

is unlikely that the incremental strain in the U-bolt w i l l be an order of magnitude greater

than that found in the preliminary measurements. Thus the next stage in development of

the U-bolt load transducer is to build a prototype that takes advantage of higher resistance

strain gauges and amplifiers to produce a reliable signal.

55
 REFERENCES

A N S Y S , 2002. Reference Manual A N S Y S Version 8.0. A N S Y S Inc., Pittsburgh, P A .

A S M , 1999. Metals Reference Book. 2 n d

ed. American Society of Metals International
(Metals Park, OH).

Burke, A . M . and O. A . Olatunbosun, 1997. "Static tyre/road interaction modelling. "

Meccanica journal, V o l . 32, pp 473-479,

British Columbia Ministry of Forests, 1999. Forest Service Bridge Design and
Construction Manual. B C M O F . . Victoria, BC (52p). Available online
"http://www.for.gov.bc.ca/hth/engineering/documents/publications_guidebooks/manuals
_standards/bridge_manual.pdf' last visited 03-11-2005.

Diamantoudis, A . Th. and Ch. A . Apostolopoulas, 2002. " B o d y mounting and A D R

requirement for tank vehicles carrying dangerous goods", Tech. Chron. Sci. Journal, V o l .
4, N o . 1-2, pp 19-25 (in Greek, extended summary in English).

Duffy, D. P., 2003. "What you should know about weighing systems." MSW
Management magazine, V o l . 13 Issue, N o . 7 N o v / D e c 2003.

Dayton Parts Ltd., 2001. U-bolts Training and Technical Manual. Product Emphasis
Program (PEP) N O . 8 (4p).
Available online "http://www.daytonparts.com/_pdf/PEP08_U-Bolts.pdf' last visited
03-11-2005.

Imaoka, S., 2000. "Modelling preloaded bolts". Memo from Collaborative Solution Inc,
(1 lp). Available online
" h t t p : / / A N S Y S . n e t / A N S Y S / t i p s / w e e k l 2-STI43_TNT_Bolt_preload.pdf' last visited
11-02-06.

K i r b y , D. and R. Charniga, 2005. " A finite element and experimental analysis of a light
truck leaf spring system subjected to pre-tension and twist loads." S A E Technical Paper
2005-01-3568, Society of Automotive Engineers. Warrendale, P A . (lOp).

Lathan, C , 1999. Construction Mechanics Basic Volume 2. U S Naval Education and

Training Professional Development and Technology Center.

Laursen, T . A . and J.C. Simo, 1993, "Algorithmic Symmetrization of Coulomb Frictional

Problems Using Augmented Lagrangians", Computers Methods in A p p l i e d Mechanics
and Engineering, V o l . 108, N o . 1 & 2, pp 133-146

Newton, W . H . , 1987. Trials of Three On-board A x l e Weighing Systems for Heavy

Goods. Research Report 103, Transport and Road Research Laboratory, U K . (24p).

56
Norton, R.L, 1996. Machine Design: A n Integrated Approach. Prentice-Hall, N e w Jersey.

Parker, S., 2004. "Hauling safety on steep road grades in British Columbia" in
Proceedings I U F R O 12 International Mountain Logging Conference. Vancouver, B C .
th

June 2004, ( l i p ) .

Phillips, E., 1989. On-board Weigh Scales for Logging Trucks and Loaders: an
Evaluation. F E R I C Technical Report T R - 9 1 . Vancouver, B C . (26p).

Oakley, P. and N . G . Marshall, 1989. Optimal Sizing of Off-highway Logging Trucks.

F E R I C Technical Report TR-96. Vancouver, B C . (20p).

Ragab, A . R . and S. E. Bayoumi, 1999. Engineering Solid Mechanics: Fundamentals and

Applications. C R C Press. Florida.

Rizzo, A . R . , 1991. " F E A Gap Element: choosing the right stiffness." Mechanical
Engineering, V o l . 113, N o 6, pp 57-59.

S A E , 1999. Manual on Design and Application of Leaf Springs, 4 th

edition. Society of
Automotive Engineers. Warrendale, P A . (122 p).

Saravi, A . and K. C. Lyons, 2004. "Finite element modelling of guyed backspars in cable
logging." In Canadian Journal of Forest Research V o l . 34, pp 817- 828.

Speck, J . A . , 1997. Mechanical Fastening, Joining and Assembly. Marcel Dekker Inc..
N e w York, N Y .

Stalling, J . M . and D. Y . Hwang, 1992. "Modelling pretensions i n bolted connections."

Computers and Structures, V o l . 45, N o . 4, pp 801-803.

Walsh, R. A . , 2000. Electromechanical design handbook. M c G r a w H i l l , N e w York, N Y .

Waugh, A . , 2005. Personal Communication, Hayes Forest Service.

Window, A . L., 1992. Strain Gauge Technology, 2 n d

edition. Elsevier. Essex, U K .

Workers Compensation Board of British Columbia, 2003. Incident Bulletin

IB2003NOA399. Vancouver, B C . (2p). Available online.
"http://publications.healthandsafetycentre.org/PDFs/SafetyBulletins/IB2003NOA399.pdf
" last visited 03-11-2005.

57
 APPENDIX I

Strain transformation along the curved part of U-bolt

Local

Shank portion Curved portion

Global x
Figure A.I: Strain transformation along the curved portion of the U-bolt

If s , s , e , y , Y
x y z xy y z and y z x are the three dimensional strain components aligned

with x y z coordinate system then the strain component e with respect to node aligned x

along with the xyz coordinate system (Ragab and Bayoumi, 1999) is

£
x =J
£ 2
+ e m + s n + y lm + y mn + y nl
y
2
z
2
yz 2X

where s , s , e are normal strain

x y z

Yxy, Yyz and Yzx are engineering shear strain

1, m and n are the directional cosine vector which was calculated from the unit

tangent vector in A N S Y S .

58
Note: A N S Y S reports the engineering shear strain which is twice the tensor shear strain

r
dv du^
y xy — + - \ = 2s r

dx dy xy

dw dv
2e
V
dy dz
r
du dw
r, = 2e.
y dz dy j

where u,v,w are the displacements at the x,y,z direction respectively

s ,s ,s
xy yz zx are the tensor shear strain

59
 A P P E N D I X II

The dimension of the slip-on water tank used in testing

All the Dimensions

are in metres

M3 Plate thickness of
CD 0.01m

FiotitViav

3£5

Top viav

Figure A.II: Slip-on water tank dimensions

60